China Prior Art Research

July 3, 2017 Miao Li

Prior Art Research in Chinese Patents

According to WIPO statistics, there were 2.9 million patent applications in 2015, with the Chinese Patent Office (SIPO) achieving the landmark of more than a million filings.[1] Indeed over the past few years, China has overshadowed its counterparts with the exponential growth in the number of its patent filings. For that reason, Chinese patent publications cannot be ignored in prior art research, regardless of whether patent protection is sought in China.[2]  This article therefore addresses the need for prior art research in Chinese patents in terms of both quantity and quality. It also considers how to run prior art search in Chinese patents, given that most people do not speak the language.

  1. Quantity and Quality of Chinese Patent Publications

Fig. 1 shows the number of Chinese patent publications each year since 1985, when the first Chinese Patent Law came into force. The number has grown exponentially since 2011. 

                          

Fig. 1

In Fig. 1, publications are divided into three types: patent application publications, patent grant publications and utility model publications. In practice, Chinese utility models have generally been regarded as being of poor quality, because there is no substantial examination for utility models and the level of inventiveness required is lower compared to invention patents. However, if we study the trend over the past two years (2015 and 2016), the gap between patent applications and utility models has widened. This indicates that applicants are shifting

from filing utility models to filing more invention patents, partly due to technical advances in China, as well as the shifting of tax incentives and policy considerations.[3] Nevertheless, even considering only granted invention patents, the total number of over 2.2 million thereof means a huge pool of prior art that no one can ignore. In many cases, the claims of Chinese invention patents have actually been found to be narrower in scope than their equivalents, i.e. Chinese granted invention patents tend to be stronger in that they are likely to be valid.[4]

Moreover, we would like to view the quality of Chinese patents from a different angle: backward citation. It is widely accepted that a backward citation has a relatively high value in terms of its technical contribution to the industry. By studying how many Chinese publications have been backward cited in a search report, we can therefore acquire a sense of the importance of Chinese references in prior art research. According to statistics, around 11 million Chinese publications (the exact number being 10,819,013) have been backward cited. However, we assume that the same document is sometimes cited multiple times in different search reports. Thus if these ‘duplicates’ are removed, around 3.9 million (the exact number being 3,862,917) are left.  Nevertheless, 3.9 million is a huge number for backward citation. [5]

In view of the rationale above, we will not overlook Chinese publications in prior art research.  To avoid missing material prior art references, you must know what has previously been published or granted in your area of expertise in China. 

  1. Translation

Having explained the necessity for prior art research in Chinese patents, the next question is obviously how to run the search. There is surely no need for everyone to understand Chinese: machine translation is generally available for Chinese patents, so we can run a search in English translation. However, with regard to such a translation, scale, dimension and timing of the translation are critical criteria to consider.

First of all, we must consider the scale of the data available in translation. There are currently over 14 million Chinese publications (invention patents and utility models). Furthermore, as mentioned above, 3.9 million have been backward cited. In this sense, the translation provided will be of an order of magnitude of millions. Human translation is almost impossible at this order of magnitude, so only machine translation is feasible. 

It could be argued that some of the references must already have an English translation, for example those with a US equivalent. That is true. We further assessed the country of the applicant of the backward cited Chinese references. The top 10 countries are listed in Table 1, where China is given 444,321 times and the other countries are far behind.[6]  Thus more than 92% of the Chinese references that have been backward cited were filed by Chinese applicants. A majority of these references probably do not have a foreign equivalent since Chinese applicants still only tend to file domestically. This means that they all require a translation.

Country of applicant of backward citations

Count

CN

440321

RU

12531

DE

4513

JP

3265

KR

2595

VG

2200

FR

1964

KY

1435

ES

1224

US

1032

 

Table 1

 

The second criterion is the dimension or field of the translation: full text translation or just bibliographic and abstract data. Accessible full text is undoubtedly important to prior art search because a short description in any part of the specification may render a pending application not novel or inventive. By contrast, a standard abstract drafted by an experienced attorney may have intentionally omitted important technical features of the invention. That explains the significance of a full text translation.

Last but not least, the timing of obtaining the translation is critical. When running a prior art search, it is desirable to get the most up-to-date references, thereby enabling a comprehensive patentability analysis. However, there is definitely a lag between the publication in the original language (e.g., Chinese) and the available date of the respective translation. The length of this time lag is therefore an important consideration for data quality. For example, a specific criterion might be the percentage of available translation within 30 days of publication in the original language. Normally a percentage of 98% or higher indicates good data quality in timing.

To summarise, in an era when machine translation is generally available, we should assess translation quality more rigorously. There are at least three criteria to consider: scale of the data available in translation, dimension of the translation and the timing of obtaining the translation.  Together, the three criteria facilitate a comprehensive yet precise prior art search for references written in languages the searcher may not understand.

Author: Miao Li, Trainer and On Boarding Manager LexisNexis IP Solutions

 


[1] http://www.wipo.int/pressroom/en/articles/2016/article_0017.html

[2] Needless to say, China is becoming an attractive jurisdiction for foreign companies to seek patent protection. There are several reasons for this, such as China becoming a more important market, the specialised IP courts set up in 2014 and the recently revised Patent Examination Guidelines that strengthen protection for software and business method inventions.  Nevertheless, this is beyond the scope of the present article, which will focus on the need for prior art research in Chinese patents.

[4] Ibid.

[5] All the data regarding backward citation in this article is as of 17 March 2017.

[6] Disclaimer: because of the low coverage of this specific data element, namely the country of the applicant, one should be careful in interpreting the figures and the ranking.

Previous Article
How to Win Patent Litigation With Big Data - Huawei vs Samsung
How to Win Patent Litigation With Big Data - Huawei vs Samsung

Ho to win patent litigation with the help of big data analytics. Analysing the case Huawei vs Samsung

Next Article
Nervous New Patent Examiners · Better Patents Now Podcast · Ep. 11
Nervous New Patent Examiners · Better Patents Now Podcast · Ep. 11

Need additional information?

Contact Us